Public Abstract

DE-SC0025277: Adjoint Methods for Advanced Accelerator Design

Award Status: Active

Institution: University of Maryland, College Park, MD
UEI: NPU8ULVAAS23
DUNS: 790934285

Most Recent Award Date: 08/12/2024
Number of Support Periods: 1
PM: Colby, Eric

Current Budget Period: 09/01/2024 - 08/31/2025
Current Project Period: 09/01/2024 - 08/31/2027
PI: Antonsen, Thomas

Supplement Budget Period: N/A

Public Abstract

Adjoint Methods for Advanced Accelerator Design Thomas Antonsen, University of Maryland, (Principal Investigator) Objective: To develop and implement efficient numerical algorithms to evaluate and improve beam quality by efficiently optimizing the design and operational adjustments of charged particle accelerators. The proposed research addresses physics that is relevant to virtually all accelerators in which a charged particle beam is confined and guided by a system of magnets, known as a lattice, that is characterized by a large number of parameters. The goal is to develop so-called “adjoint” methods that efficiently determine the sensitivity of various beam quality metrics to all the parameters describing the transport lattice. Knowledge of the sensitivity can be used in two ways. It can be used to establish manufacturing tolerances for the magnet parameters, and it can be used as part of a design scheme in which parameters are varied to optimize a design. To be useful, the adjoint calculation needs to be implemented in a simulation code. We will do this in collaboration with the developers of the code OPAL, which is freely available to the community. The proposed research will impact DoE efforts in both Theory and Accelerator Science and Technology R&D. It will also serve to educate students in computational accelerator science. The traditional approach to determining the sensitivity of an accelerator lattice performance to the values of the parameters characterizing the lattice, for example the strengths, orientations, and locations of the magnets, is to individually vary each parameter and tabulate the change in performance. This requires simulating the beam evolution in the lattice as many times as there are parameters to vary. With the adjoint approach, the sensitivity of the figure of merit to all parameters can be found by one or two, “time-reversed” runs of the code used to model the system. So that our methodology will impact the designs of future accelerators, we will implement it in a freely available simulation code and provide to the community examples of its use. It will then be readily adaptable to accelerators in high-energy physics, medical accelerators, light sources, spallation neutron sources, and heavy ion storage rings such as those at Fermi Lab or GSI-Darmstadt in Germany. A central component of our mission is educating the next generation of researchers. The University of Maryland has been a leader in graduate education within the field: our group has produced 10 Ph.D. graduates in charged particle beam research in the last decade, and dozens more in prior decades, many of whom now work at national laboratories and in industry. We have also awarded many Masters degrees, and trained numerous undergraduate and high school students, providing paid and volunteer research opportunities through internships and independent study projects. A recent count tallied dozens of undergraduate and high school students who performed research with our group in the last 10 years, many of whom have gone on to pursue college degrees and careers in STEM fields. Research will be conducted in the Institute for Research in Electronics and Applied Physics, which has established a task force, IREAP-ROLE (https://ireap.umd.edu/ireap-role) to promote diversity, equity and inclusion. The IREAP task force also coordinates with a similar effort in the Physics Department.

Adjoint Methods for Advanced Accelerator Design

Thomas Antonsen, University of Maryland, (Principal Investigator)

Objective: To develop and implement efficient numerical algorithms to evaluate and improve beam quality by efficiently optimizing the design and operational adjustments of charged particle accelerators. The proposed research addresses physics that is relevant to virtually all accelerators in which a charged particle beam is confined and guided by a system of magnets, known as a lattice, that is characterized by a large number of parameters. The goal is to develop so-called “adjoint” methods that efficiently determine the sensitivity of various beam quality metrics to all the parameters describing the transport lattice. Knowledge of the sensitivity can be used in two ways. It can be used to establish manufacturing tolerances for the magnet parameters, and it can be used as part of a design scheme in which parameters are varied to optimize a design. To be useful, the adjoint calculation needs to be implemented in a simulation code. We will do this in collaboration with the developers of the code OPAL, which is freely available to the community. The proposed research will impact DoE efforts in both Theory and Accelerator Science and Technology R&D. It will also serve to educate students in computational accelerator science.

The traditional approach to determining the sensitivity of an accelerator lattice performance to the values of the parameters characterizing the lattice, for example the strengths, orientations, and locations of the magnets, is to individually vary each parameter and tabulate the change in performance. This requires simulating the beam evolution in the lattice as many times as there are parameters to vary. With the adjoint approach, the sensitivity of the figure of merit to all parameters can be found by one or two, “time-reversed” runs of the code used to model the system. So that our methodology will impact the designs of future accelerators, we will implement it in a freely available simulation code and provide to the community examples of its use. It will then be readily adaptable to accelerators in high-energy physics, medical accelerators, light sources, spallation neutron sources, and heavy ion storage rings such as those at Fermi Lab or GSI-Darmstadt in Germany.

A central component of our mission is educating the next generation of researchers. The University of Maryland has been a leader in graduate education within the field: our group has produced 10 Ph.D. graduates in charged particle beam research in the last decade, and dozens more in prior decades, many of whom now work at national laboratories and in industry. We have also awarded many Masters degrees, and trained numerous undergraduate and high school students, providing paid and volunteer research opportunities through internships and independent study projects. A recent count tallied dozens of undergraduate and high school students who performed research with our group in the last 10 years, many of whom have gone on to pursue college degrees and careers in STEM fields. Research will be conducted in the Institute for Research in Electronics and Applied Physics, which has established a task force, IREAP-ROLE (https://ireap.umd.edu/ireap-role) to promote diversity, equity and inclusion. The IREAP task force also coordinates with a similar effort in the Physics Department.